You very deeply do not understand the Apple vs. Microsoft lawsuit. That lawsuit had nothing to do with broad technologies, as you seem to fear, but rather with very specific behaviors of a few icons and the actions they performed. I guess it's too much to expect an anonymous coward to actually learn about something before spouting off about it though.
Claiming Open Source is a threat to intellectual property. This manages to insult open source people, suggesting they're to stupid to understand the implications of open source.
Claiming Open Source is a threat to intellectual property also assumes that people in our industry are hot for intellectual property laws - when many of us are seriously rethinking them. Mr. Mundie somehow assumes we're all stupid sheep who love the current intellectual property system.
And some of us are not rethinking them, and realize that the GPL is a threat to intellectual property. In point of fact, most of what he said/wrote is true. Just because you don't agree with him doesn't make him a liar, it just makes him disagree with you. Believe it or not, your opinion is not fact.
The swipe at IBM merely insults one of their competitors' intelligence - a competitor that already is likely to not feel particuarly merciful.
So? Standard marketing tactic for every company, including IBM. The implication of every marketing campaign is ``our stuff is so much better (cheaper, faster) than our competitors they're stupid for making it and you're stupid for using it.'' This pretty much includes the marketing for the GPL and for Linux. Again, disagreeing with you doesn't make him a liar.
Open source has a forking-software problem. Try the joy of dealing with several different versions of Windows in the same office, hardware and driver issues, ad nauseum. He assumes we forget some of the weird stuff Microsoft has pulled.
Now this was funny. After the Kerberos fiasco, it took giant brass balls to say this.
The GPL is not understood by many sophisticated people. Yeah, like it represents all of Open Source -- and exactly how is he so sure those of us who use the GPL don't understand it? Again, he literally tells us we're idiots.
The GPL is not understood by people, sophisticated or otherwise. The GPL doesn't really mean anything until it has been interpreted in court, which hasn't happened -- yet. When it does, I'm pretty much betting everyone is going to be surprised. It is impossible to predict what interpretation might happen, due to the imprecise language of the GPL. In particular, the phrase ``when you distribute the same sections as part of a whole which is a work based on the Program, the distribution of the whole must be on the terms of this License'' is going to give anyone putting GPL code into an embedded product a real nightmare in court. (GPL v2, Section 2, paragraph after item c).
"VA can call itself whatever it wants to, but under the US Constitution it's a state".
Yes, and if they were trying to get the license interpreted under US law, that would probably be effective. They're not, it won't work, and you're a moron for not noticing that. Try not to dabble in legal work, you obviously don't have the powers of concentration needed.
Uh, yeah, oddly enough this idea *does* work. See, for instance, http://www.cogolink.net/
I am a customer, as is my company and 3 co-workers. We run a VPN over the wireless network so we can comfortably work from home. Of course, most of their other customers are brainless Windows lusers with unprotected file sharing, but we sure get great throughput outside the 9-5 moron window.
In case you're wondering, the residential service provides bandwidth-limited internet access, but the bandwidth limiting is done at the ISP. Internal communications over the wireless network run at whatever speed the network is currently capable of. Typical speeds in the evening are in the 2.5 - 3.0 mbit/s range, with latency between stations of about 10 - 15 ms.
These guys run on an all-wireless network, with access points scattered all over the Ogden, Salt Lake City, and Provo area. The access points are connected to each other via 802.11 links also. The network is, of course, sensitive to rain and snow, but is quite reliable. Most of the service outages I've experienced in the past year have been human (operations) rather than weather or technology failures.
Some fool named CAILAS penned: things seem to be getting tighter and tighter, as the 'bad distributions' get weeded out and kernel 2.4 was (somewhat) recently released, bringing it quite near to par with the BSD kernel
And they said there wasn't anything funny in this feature. See, nerds can be funny when they don't try.
Not to nitpick or anything, but...
I hear this all the time. OS X is NOT "just BSD with some apple stuff slapped on top." It's not even truly BSD. It's a Mach Microkernel with a BSD Compatibility layer on top.
If you're going to pick nits, you ought to at least know what you're talking about before you start picking.
OS X uses a BSD kernel, based on NetBSD code, running on a Mach microkernel. It's not a "BSD emulator", it's a BSD kernel on top of the microkernel. If you don't understand the difference, you certainly shouldn't be shuffling your feet in and out of your mouth, it can't sustatin that I/O rate.
The OS X userland code is based closely on a recent FreeBSD distribution. Apple has, of course, moved a few things around, extended a few things here and there, and made the BSD system fit into their needs, but the system is a BSD system.
I just exchanged email with Kristin Ford at Evans Data; the FreeBSD data is being restored to the site. They will also include the table with the "Linux only" stuff for the crybabies...
Bugtraq and the Netcraft uptime report don't agree with you that the difference between BSD and Linux is "just religion", unless your religion is staying up and unhacked.
I didn't notice any mention of the underlying FreeBSD based server achitecture during the broadcast today on TechTV. I do wonder how much server will remain behind the client-centric GUI. Will this be something to replace our Linux/XFree setup?
It's all still in there. Jordan K. Hubbard was recently interviewed and commented how sublimed it was to see GNU Emacs running on a Mac. Apache builds and installs -- I think it's included in the Public Beta. XF86 4.02 compiles on Darwin, the Open Source version of the operating system (minus the stunning Aqua user interface).
As OS X gets out and forms more of a developer community, bring over UNIX applications from FreeBSD (and Linux) will advance. Imagine an inexpensive, reliable computer like the iMac with the lovely GUI for family members and FreeBSD's 4500+ UNIX applications, stellar networking, and a reliable OS under the hood.
I already got an iMac for my father, maybe it's time to get me one too.
I agree. In point of fact, VA Research used to sell (and preinstall, I believe) FreeBSD on their systems, but stopped about the time they became VA Linux Systems. This differs from Compaq or Dell offering nothing but MS trash on their systems not one iota, and claims of "VA Linux" being an Open Systems company are marketing lies. They walked away from their Open Source stance to become VA Linux.
Your comment doesn't piss me off, and I'm a pretty hard-core BSD "folk." I think you've provided a fairly accurate overview of the difference in the two communities here.
There are a lot of people who do use BSD for workstations out there, too. BSD even leads in some areas, with support for USB well in advance of Linux, and active FreeBSD and NetBSD members on many of the Open Source application teams now. This is an area in which BSD is catching up to Linux, though.
Er, no, they DID pick that OEM partly because it runs FreeBSD. This product was picked largely due to technical forethought. I worked for the company that became ``Intel InBusiness'' and know most of the engineering staff quite well.
Keep an eye on Dæmon News for more information about this issue, if you care to have your prejudices rearranged.
yer NOT gunna need a cheat sheet
on
Which BSD?
·
· Score: 1
I, on the other tentacle, recommend getting The Complete FreeBSD by Greg Lehey with the FreeBSD CD-ROM set and starting from there. It is also a great book; Tim O'Reilly has publicly lamented deciding not to publish it. The Complete FreeBSD will walk you through the installation of FreeBSD and then provide help in configuring FreeBSD to do a variety of work, including various internet services and X windows workstation configuration.
Both FreeBSD and Linux have IP stacks so fast they are capable of easily overrunning slow Win9x machines, causing the Win9x machines to drop packets. This is a good indication both are using the hardware pretty efficiently.
The Linux TCP/IP implementation is different because that's how the Linux gang wanted it to be; they wanted to develop a GPL'd IP stack and they have done a remarkable job. They couldn't just incorporate the BSD networking stack because the license on the BSD code would not allow it to be GPL'd.
I'd sum up the performance of both as being darned close to "as good as it gets."
Re:Just what the hell is that supposed to mean?
on
Which BSD?
·
· Score: 2
Linux is the "WinTel hardware world"?!?
Just off the top of my head, I can think of several platforms that Linux runs on:
PPC
x86
ARM
SPARC
Alpha
You BSD guys just never give it a rest, do you?
Add to that list
VAX
Sun-3
HP 300/400
Mac 68K
Amiga
Atari TT/Falcon
NeXT
MIPS HHPC
NS32532
BeBox
Hitachi SH3
But don't take my word for it, go visit the NetBSD Supported Hardware page and see for yourself. Nobody implied Linux isn't portable, just that NetBSD has been ported to more platforms than just about anything else. Next time, try to keep your knee from jerking quite so hard.
Re:Linux emulation far from perfect--no there yet
on
Which BSD?
·
· Score: 1
Yeah, yeah, yeah, more Anonymous Coward drivel. If you have something to say, say it with your name on it.
FreeBSD has support for all Posix thread functions except thread_cancel. It has had this level of support since version 3.0. Most "non-trivial" Linux binaries, such as WordPerfect, Wingz, StarOffice, etc., require a few simple tweaks to the installation script so it will recognize FreeBSD as a "variant" of Linux. These are provided by FreeBSD volunteers who create "port" kits that tweak the scripts and run the install program.
Pardon me, Anonymous Coward, but I believe YOUR agenda is showing.
Well, there were a lot fewer BSD-related articles than got submitted. I personally submitted at least a dozen different articles this year. Total that made it to/.: 0. Zip. Nil. Yeah, the BSD articles were *really* getting through.
If you really want to learn what's going on in the BSD world, go have a look at Dæmon News Daily, where they print all the BSD news, not just what gets past the/. censor corps.
Reading this column, I just had an epiphany. As the author went round the bend from analyzing what the latest giant hole in HotMail meant to how this proves Open Source is an extinction-level event for Microsoft, I finally realized an important point I have been groping for the past several years.
I can now characterize the primary difference between Linux zealots and BSD zealots in one simple phrase:
Linux zealots are firmly convinced that Open Source software is going to save
everyone, while BSD zealots only care about saving the nerds.
Think about it: this simple difference in viewpoint encompasses the differences between the developer communities, the user communities, and even the hallmark licenses of both camps: the GPL which vests ownership of the code in ``the community'' vs. the BSD license vests ownership of the code with anyone who wants to use it.
It is hard to say which will prove to have the longest lasting effect on the world at this point. I have a pretty jaded viewpoint on how much John Q. Public wants to be saved from the evil that lurks within his phone, television, or internet connection, so long as he can figure out how to use it by watching a video that is not more than 1 commercial break long.
As for my house, we will stick with the nerds. I'm too busy to save the world. Even from themselves.
Simoriah said ``I WOULD have liked to see more info comparing the different flavors of xBSD, though. ''
Go back to Daemon News and read some of the other columns and articles, or maybe a few back issues. This isn't the only thing that has ever appeared there, you know.;^)
Let me reply to a few of your comments. First, you say ``He trivializes Linux developers for writing drivers''. No, I did not, in fact I lauded them for their efforts to support every piece of hardware the PC platform can throw at them. I remain amazed at the incredible ability of Linux device driver writers to reverse engineer some of the most bizarre and stupid hardware on the planet, and applaud them for this effort. You're trying to put words into my article, but oddly enough they still haven't appeared in it.
``I don't see a Code Warrior for xBSD''. I don't see a Code Warrior for Debian, or SuSE, or Caldera, or SlackWare, or Turbo, or any other Linux-based operating systems either. I do happen to have a beta of Code Warrior for Solaris at work, and like Code Warrior for RedHat, it's not terribly impressive. It's a half-baked editor with a cheesy binary project file build system wrapped around the same GNU compiler you can download for free from Cygnus. I'll stick with Emacs, thank you. If you want to look at an innovative approach to building complicated systems, see Jam/MR from Perforce Software. It's available under a free license and runs on Linux, xBSD, and just about anything else. Before you jump to point out that CodeWarrior for RedHat will run on Debian, or SuSE, or whatever, let me assure you it will run on FreeBSD as well.
``Linux is not xBSD on training wheels. It's an alternative, original implementation of classic UNIX concepts.'' Yup, you're exactly right. What my article points out is that Linux and xBSD are very closely related, both by the direct cross-pollination between the development groups and by the amount of shared code they use. Both Greg Lehey and I have written before about the difference in focus between FreeBSD, OpenBSD, and NetBSD, and I was showing our mostly BSD-based readership that Linux is very similar. In point of fact, it is probably the most like NetBSD, due to the portability of the Linux kernel, but is fairly close to FreeBSD as well, since much of the focus remains on the PC architecture. In addition to that, it has additional focuses, such as embracing a wide range of hardware, that differentiate it from all of the BSDs.
I suggested Linux is the best starting place for those who don't already know UNIX not because Linux is some goofball stripped down system with ``training wheels'', but rather because the developers of Linux have done such a good job of making it run and run well on just about anything. The developers of Linux also really care about those guys with the $29 tape drive, because they were all ``some guy with a $29 tape drive'' once themselves.
So, let's stop reading all your Linux snobbery into my article and take it at face value. When I say developing a device driver for a $29 tape drive doesn't mean the developer is wasting his time, that is exactly what I mean. I phrased it the way I did because that was the next natural question for the Horshacks in the audience to raise: ``Oooh! Oooh! Mr. Kotter! Doesn't that mean the developer is like wasting his time?'' ``No, Arnold, it means he has a working tape drive and you do too.''
Some Anonymous Coward wrote ``Your benchmark is bogus. Use netperf. You'll see that both systems saturate the Ethernet in any combinaisons [sic] (albeit with some minor differences).''
Netperf is a tool for measuring network throughput, but has nothing to do with FTP throughput. If you want to measure FTP throughput, you have to use actual FTP servers and clients. A more fair test would be to install FreeBSD and Linux on both machines and test each combination of FreeBSD and Linux, server and client, running on each machine, to see if you can characterize performance. Short of that, you have a few datapoints but not enough to draw any conclusions.
Some of the FreeBSD ethernet drivers do actually work fast enough to overrun anything but a full duplex switched network. The time between packet transmissions is so short it has been known to run Windows ftp clients straight into the ground, stuffing data into them faster than they can take it. Linux, on the other hand, is capable of receiving these packets quickly enough to not require retries.
haapi writes ``It would be nice to see the recent Mindcraft benchmarks run against *BSD. The MS guys wouldn't have to change a thing -- just bring in the *BSD experts and have them tune-up their machines and go...''
They have been, and Solaris as well. Guess which one came out on top, beating even NT? Solaris, of course. Anyone who thinks those Sun guys and gals sleep for a living is deluded.
Like the Linux engineers participating in the test, the FreeBSD people participating learned a lot about sustaining high througput on a high-end server, and learned a lot about how the system performs currently and how it might be improved. I don't have specifics on the performance numbers, nor when they might be published. I doubt the numbers I heard had even been audited for accuracy yet, but I suspect that both Linux and FreeBSD will perform much better in the next round. Both have already seen development (in Linux 2.3 and FreeBSD 4.0) based on what was learned at this benchmark, and we'll all benefit from this.
It'll be interesting to see if Microsoft can keep up.
Slashdot Mirror
You very deeply do not understand the Apple vs. Microsoft lawsuit. That lawsuit had nothing to do with broad technologies, as you seem to fear, but rather with very specific behaviors of a few icons and the actions they performed. I guess it's too much to expect an anonymous coward to actually learn about something before spouting off about it though.
-
Claiming Open Source is a threat to intellectual property. This manages to insult open source people, suggesting they're to stupid to understand the implications of open source.
- Claiming Open Source is a threat to intellectual property also assumes that people in our industry are hot for intellectual property laws - when many of us are seriously rethinking them. Mr. Mundie somehow assumes we're all stupid sheep who love the current intellectual property system.
And some of us are not rethinking them, and realize that the GPL is a threat to intellectual property. In point of fact, most of what he said/wrote is true. Just because you don't agree with him doesn't make him a liar, it just makes him disagree with you. Believe it or not, your opinion is not fact.- The swipe at IBM merely insults one of their competitors' intelligence - a competitor that already is likely to not feel particuarly merciful.
So? Standard marketing tactic for every company, including IBM. The implication of every marketing campaign is ``our stuff is so much better (cheaper, faster) than our competitors they're stupid for making it and you're stupid for using it.'' This pretty much includes the marketing for the GPL and for Linux. Again, disagreeing with you doesn't make him a liar.- Open source has a forking-software problem. Try the joy of dealing with several different versions of Windows in the same office, hardware and driver issues, ad nauseum. He assumes we forget some of the weird stuff Microsoft has pulled.
Now this was funny. After the Kerberos fiasco, it took giant brass balls to say this.- The GPL is not understood by many sophisticated people. Yeah, like it represents all of Open Source -- and exactly how is he so sure those of us who use the GPL don't understand it? Again, he literally tells us we're idiots.
The GPL is not understood by people, sophisticated or otherwise. The GPL doesn't really mean anything until it has been interpreted in court, which hasn't happened -- yet. When it does, I'm pretty much betting everyone is going to be surprised. It is impossible to predict what interpretation might happen, due to the imprecise language of the GPL. In particular, the phrase ``when you distribute the same sections as part of a whole which is a work based on the Program, the distribution of the whole must be on the terms of this License'' is going to give anyone putting GPL code into an embedded product a real nightmare in court. (GPL v2, Section 2, paragraph after item c).Yes, and if they were trying to get the license interpreted under US law, that would probably be effective. They're not, it won't work, and you're a moron for not noticing that. Try not to dabble in legal work, you obviously don't have the powers of concentration needed.
I am a customer, as is my company and 3 co-workers. We run a VPN over the wireless network so we can comfortably work from home. Of course, most of their other customers are brainless Windows lusers with unprotected file sharing, but we sure get great throughput outside the 9-5 moron window.
In case you're wondering, the residential service provides bandwidth-limited internet access, but the bandwidth limiting is done at the ISP. Internal communications over the wireless network run at whatever speed the network is currently capable of. Typical speeds in the evening are in the 2.5 - 3.0 mbit/s range, with latency between stations of about 10 - 15 ms.
These guys run on an all-wireless network, with access points scattered all over the Ogden, Salt Lake City, and Provo area. The access points are connected to each other via 802.11 links also. The network is, of course, sensitive to rain and snow, but is quite reliable. Most of the service outages I've experienced in the past year have been human (operations) rather than weather or technology failures.
things seem to be getting tighter and tighter, as the 'bad distributions' get weeded out
and
kernel 2.4 was (somewhat) recently released, bringing it quite near to par with the BSD kernel
And they said there wasn't anything funny in this feature. See, nerds can be funny when they don't try.
If you're going to pick nits, you ought to at least know what you're talking about before you start picking.
OS X uses a BSD kernel, based on NetBSD code, running on a Mach microkernel. It's not a "BSD emulator", it's a BSD kernel on top of the microkernel. If you don't understand the difference, you certainly shouldn't be shuffling your feet in and out of your mouth, it can't sustatin that I/O rate.
The OS X userland code is based closely on a recent FreeBSD distribution. Apple has, of course, moved a few things around, extended a few things here and there, and made the BSD system fit into their needs, but the system is a BSD system.
I just exchanged email with Kristin Ford at Evans Data; the FreeBSD data is being restored to the site. They will also include the table with the "Linux only" stuff for the crybabies...
Bugtraq and the Netcraft uptime report don't agree with you that the difference between BSD and Linux is "just religion", unless your religion is staying up and unhacked.
It's all still in there. Jordan K. Hubbard was recently interviewed and commented how sublimed it was to see GNU Emacs running on a Mac. Apache builds and installs -- I think it's included in the Public Beta. XF86 4.02 compiles on Darwin, the Open Source version of the operating system (minus the stunning Aqua user interface).
As OS X gets out and forms more of a developer community, bring over UNIX applications from FreeBSD (and Linux) will advance. Imagine an inexpensive, reliable computer like the iMac with the lovely GUI for family members and FreeBSD's 4500+ UNIX applications, stellar networking, and a reliable OS under the hood.
I already got an iMac for my father, maybe it's time to get me one too.
Nobody in their right mind runs X on a server, and it certainly isn't going to do you any good trying to figure out why the remote machine won't boot.
Next time, try reading the linked article.
I agree. In point of fact, VA Research used to sell (and preinstall, I believe) FreeBSD on their systems, but stopped about the time they became VA Linux Systems. This differs from Compaq or Dell offering nothing but MS trash on their systems not one iota, and claims of "VA Linux" being an Open Systems company are marketing lies. They walked away from their Open Source stance to become VA Linux.
There are a lot of people who do use BSD for workstations out there, too. BSD even leads in some areas, with support for USB well in advance of Linux, and active FreeBSD and NetBSD members on many of the Open Source application teams now. This is an area in which BSD is catching up to Linux, though.
If you ask nicely, they'll even pre-load FreeBSD on it for you. ;^)
Keep an eye on Dæmon News for more information about this issue, if you care to have your prejudices rearranged.
I, on the other tentacle, recommend getting The Complete FreeBSD by Greg Lehey with the FreeBSD CD-ROM set and starting from there. It is also a great book; Tim O'Reilly has publicly lamented deciding not to publish it. The Complete FreeBSD will walk you through the installation of FreeBSD and then provide help in configuring FreeBSD to do a variety of work, including various internet services and X windows workstation configuration.
The Linux TCP/IP implementation is different because that's how the Linux gang wanted it to be; they wanted to develop a GPL'd IP stack and they have done a remarkable job. They couldn't just incorporate the BSD networking stack because the license on the BSD code would not allow it to be GPL'd.
I'd sum up the performance of both as being darned close to "as good as it gets."
Just off the top of my head, I can think of several platforms that Linux runs on:
You BSD guys just never give it a rest, do you?
Add to that list
But don't take my word for it, go visit the NetBSD Supported Hardware page and see for yourself. Nobody implied Linux isn't portable, just that NetBSD has been ported to more platforms than just about anything else. Next time, try to keep your knee from jerking quite so hard.
FreeBSD has support for all Posix thread functions except thread_cancel. It has had this level of support since version 3.0. Most "non-trivial" Linux binaries, such as WordPerfect, Wingz, StarOffice, etc., require a few simple tweaks to the installation script so it will recognize FreeBSD as a "variant" of Linux. These are provided by FreeBSD volunteers who create "port" kits that tweak the scripts and run the install program.
Pardon me, Anonymous Coward, but I believe YOUR agenda is showing.
Darby rules! What a Daem.
If you really want to learn what's going on in the BSD world, go have a look at Dæmon News Daily, where they print all the BSD news, not just what gets past the /. censor corps.
I can now characterize the primary difference between Linux zealots and BSD zealots in one simple phrase:
Think about it: this simple difference in viewpoint encompasses the differences between the developer communities, the user communities, and even the hallmark licenses of both camps: the GPL which vests ownership of the code in ``the community'' vs. the BSD license vests ownership of the code with anyone who wants to use it.
It is hard to say which will prove to have the longest lasting effect on the world at this point. I have a pretty jaded viewpoint on how much John Q. Public wants to be saved from the evil that lurks within his phone, television, or internet connection, so long as he can figure out how to use it by watching a video that is not more than 1 commercial break long.
As for my house, we will stick with the nerds. I'm too busy to save the world. Even from themselves.
Go back to Daemon News and read some of the other columns and articles, or maybe a few back issues. This isn't the only thing that has ever appeared there, you know. ;^)
``I don't see a Code Warrior for xBSD''. I don't see a Code Warrior for Debian, or SuSE, or Caldera, or SlackWare, or Turbo, or any other Linux-based operating systems either. I do happen to have a beta of Code Warrior for Solaris at work, and like Code Warrior for RedHat, it's not terribly impressive. It's a half-baked editor with a cheesy binary project file build system wrapped around the same GNU compiler you can download for free from Cygnus. I'll stick with Emacs, thank you. If you want to look at an innovative approach to building complicated systems, see Jam/MR from Perforce Software. It's available under a free license and runs on Linux, xBSD, and just about anything else. Before you jump to point out that CodeWarrior for RedHat will run on Debian, or SuSE, or whatever, let me assure you it will run on FreeBSD as well.
``Linux is not xBSD on training wheels. It's an alternative, original implementation of classic UNIX concepts.'' Yup, you're exactly right. What my article points out is that Linux and xBSD are very closely related, both by the direct cross-pollination between the development groups and by the amount of shared code they use. Both Greg Lehey and I have written before about the difference in focus between FreeBSD, OpenBSD, and NetBSD, and I was showing our mostly BSD-based readership that Linux is very similar. In point of fact, it is probably the most like NetBSD, due to the portability of the Linux kernel, but is fairly close to FreeBSD as well, since much of the focus remains on the PC architecture. In addition to that, it has additional focuses, such as embracing a wide range of hardware, that differentiate it from all of the BSDs.
I suggested Linux is the best starting place for those who don't already know UNIX not because Linux is some goofball stripped down system with ``training wheels'', but rather because the developers of Linux have done such a good job of making it run and run well on just about anything. The developers of Linux also really care about those guys with the $29 tape drive, because they were all ``some guy with a $29 tape drive'' once themselves.
So, let's stop reading all your Linux snobbery into my article and take it at face value. When I say developing a device driver for a $29 tape drive doesn't mean the developer is wasting his time, that is exactly what I mean. I phrased it the way I did because that was the next natural question for the Horshacks in the audience to raise: ``Oooh! Oooh! Mr. Kotter! Doesn't that mean the developer is like wasting his time?'' ``No, Arnold, it means he has a working tape drive and you do too.''
Class dismissed.
Netperf is a tool for measuring network throughput, but has nothing to do with FTP throughput. If you want to measure FTP throughput, you have to use actual FTP servers and clients. A more fair test would be to install FreeBSD and Linux on both machines and test each combination of FreeBSD and Linux, server and client, running on each machine, to see if you can characterize performance. Short of that, you have a few datapoints but not enough to draw any conclusions.
Some of the FreeBSD ethernet drivers do actually work fast enough to overrun anything but a full duplex switched network. The time between packet transmissions is so short it has been known to run Windows ftp clients straight into the ground, stuffing data into them faster than they can take it. Linux, on the other hand, is capable of receiving these packets quickly enough to not require retries.
They have been, and Solaris as well. Guess which one came out on top, beating even NT? Solaris, of course. Anyone who thinks those Sun guys and gals sleep for a living is deluded.
Like the Linux engineers participating in the test, the FreeBSD people participating learned a lot about sustaining high througput on a high-end server, and learned a lot about how the system performs currently and how it might be improved. I don't have specifics on the performance numbers, nor when they might be published. I doubt the numbers I heard had even been audited for accuracy yet, but I suspect that both Linux and FreeBSD will perform much better in the next round. Both have already seen development (in Linux 2.3 and FreeBSD 4.0) based on what was learned at this benchmark, and we'll all benefit from this.
It'll be interesting to see if Microsoft can keep up.